Publication

Abstract :

This work introduces an adaptive and intelligent modulation scheme selection system with application to real-time wireless communications. Through the integration of Software Defined Radio (SDR) technology based on GNU Radio with machine learning (ML) algorithms, the system makes the task of signal modulation detection in adaptive and noisy situations automatic. The suggested hybrid approach consists of an offline stage for generating synthetic signals, extracting features, and training ML models, and an online stage for real-time monitoring and signal classification. Using three popular modulation schemes, namely BPSK, QPSK, and 16-QAM, the system extracts both time-domain features and spectral features to train ML classifiers and then ensembles them through majority voting. Real-time performance is evaluated using dynamic signal sampling at 15-second intervals, achieving high classification accuracy and adaptability in different signal-to-noise ratios (SNR). The outputs are displayed through a dashboard and saved in periodic reports, and therefore this system is a scalable and feasible solution for intelligent radio signal analysis in IoT and wireless systems.